Twin Peaks in ATLAS

For the annual December CERN council meeting the ATLAS experiment provided an update of the Higgs searches in the γγ and ZZ→4 leptons  channels. The most interesting thing about the HCP update a month ago was why these most sensitive channels were *not*  updated (also CMS chose not to update γγ).  Now we can see why. The ATLAS analyses in these channels return the best fit Higgs masses that differ by more than 3 GeV: 123.5 GeV for ZZ and 126.6 GeV for γγ, which is much more than the estimated resolution of about 1 GeV. The tension between these 2 results is estimated to be 2.7σ.  Apparently, ATLAS used this last month to search for the systematic errors that might be responsible for the discrepancy but, having found nothing, they decided to go public.

One may be tempted to interpret the twin peaks as 2 separate Higgs-like particles. However in this case they most likely signal a systematic problem rather than interesting physics. First, it would be quite a coincidence to have two Higgs particles so close in mass (I'm not aware of a symmetry that could ensure it). Even if the coincidence occurs, it would be highly unusual that one Higgs decays dominantly to ZZ and the other dominantly to γγ, each mimicking pretty well the standard Higgs rate in the respective channel. Finally, and most importantly, CMS does not see anything like that; actually their measurements give a reverse picture. In the ZZ→4l channel CMS measures mh=126.2±0.6 GeV,  above  (but well within the resolution) the best fit mass they find in  the γγ channel which is 125.1±0.7 GeV GeV. That makes us certain that down-to-earth reasons are responsible for the double vision in ATLAS, the likely cause being  an ECAL calibration error, an unlucky background fluctuation, or alcohol abuse. 

The truly exciting thing about the new ATLAS results is that the diphoton rate continues to be  high. Recall that we are scared as fudge that the Higgs will turn out to be the boring one predicted by the standard model, and we're desperately looking out for some non-standard behavior. The measurements of Higgs decays to ZZ and WW do not bring any consolation: all rates measured by CMS and ATLAS so far are perfectly consistent with the standard model. Today's ATLAS update in the ZZ→4l channel continues the depressing trend, with the signal strength normalized to the standard model  one   measured at 1.0±0.4 (for mh=125 GeV). Currently our best hope is that the measured h→γγ cross section is consistently larger than the one predicted  by the standard model, both in ATLAS and CMS. If the enhancement is due to a statistical fluctuation one would expect it becomes less significant as more data is added. Instead, in ATLAS,  the central value of  has not moved since July, but the error has shrunk a bit! The current diphoton signal strength stands at  1.8 ± 0.4, roughly 2 sigma above the standard model. On the other hand, given there is something weird about the ATLAS Higgs data (be it miscalibration or fluctuation),  we should treat that excess with a grain of salt, at least until the double vision problem is resolved. And we're waiting for CMS to come out with what they have in the diphoton channel...

One more news today is that ATLAS also began studying some differential observables related to the Higgs boson, which usually goes by the name of "spin determination". In particular, they looked at the production and decay angles in the ZZ→4l channel  (similar to what CMS showed at HCP) and the Higgs production angle in the γγ channel (first measurement of this kind). For spin zero the production angle  should be isotropic (at the parton level, in the center-of-mass frame of the collision) while for higher spins some directions with respect to the beam axis could be preferred.  Not surprisingly, the measured Higgs production angle is perfectly consistent with the zero spin hypothesis   (ATLAS also quotes spin-2 being disfavored at 90% confidence level, although in reality they disfavor a particular spin-2 benchmark model).

Here are the links to the ATLAS diphoton, ZZ, and combination notes.